Architecture-aware field affinity estimation

Abstract

A data layout optimization may utilize affinity estimation between paris of fields of a record in a computer program. The affinity estimation may be determined based on a trace of an execution and in view of actual processing entities performing each access to the fields. The disclosed subject matter may be configured to be aware of a specific architecture of a target computer having a plurality of processing entities, executing the program so as to provide an improved affinity estimation which may take into account both false sharing issues, spatial locality improvement and the like.

Description

BACKGROUND[0001]The present disclosure relates to performance optimizations of computer programs in general, and to field affinity estimation based on an execution by a computer having a plurality of processing entities in particular.[0002]Performance of a computer program may be improved based on performance optimizations methods. Such methods include modifying the computer program, also referred to as a target program, to perform in a more efficient manner such as for example to be executed more rapidly, to require less resources such as memory space or draw less power. Although referred to as optimizing performance of the target program, such methods do not necessarily provide the best possible performance. In some cases, performing an optimization may unfortunately lead to a reduction in efficiency of the target program in some executions or even unexpectedly in all executions.[0003]Data layout optimizations are utilized to change an individual record layout used in the target p...

AI Technical Summary

Benefits of technology

[0005]One exemplary embodiment of the disclosed subject matter is a computer-implemented method for determining affinity estimation between fields of a record in a computer program, the method comprising: obtaining a trace associated with an execution of the computer program by a computer having a plurality of processing entities, the trace comprising at least a first indication and a second indication, the first indication is associated with a first access to a first field of the record, the second indication is associated with a second access to a second field of the record, the first access precedes the second access; determining a first and a second processing entities of the plurality of processing entities associated with the first and the second indications; determining the affinity estimation based on a change in expected performance due to associating the first and second fields in a single cache line in view of performing the first access by the first processing entity and the second access by the second access entity; modifying a data layout of the computer program based on the affinity estimation.

[0006]Another exempl...

Problems solved by technology

Although referred to as optimizing performance of the target program, such methods do not necessarily provide the best possible performance.

In some cases,...

Images